Closing line value calculator
Enter the price you took and the closing prices on every side. This removes the book's margin from the close under three published methods, then turns beating it into an expected return, not just a price gap.
Worked example — Expected return per unit staked, after the vig comes out: +2.88%. Taken at +125, closed +110 with -130 the other way. Raw closing line value is +7.14%; proportional de-vig removes the 3.98% hold.
Looks like value, isn’t: The page’s own worked example: 7.14% raw closing line value that is really 2.88% after the vig comes out.
Confidence and de-vig method 1
The closing market holds 3.98%. Removing it proportionally puts the fair probability at 45.73%, a fair price of +119, so the +7.14% you beat the close by is worth +6.86% of stake, and the bet as a whole returns +2.88%.
Everything computes from the prices you type; this page reads no venue and stores no bets.
| Method | Fair probability | Fair price | Value of beating the close | Expected return per unit | Bets before it clears 95% |
|---|---|---|---|---|---|
| Proportional | 45.73% | +119 / 2.19 | +6.86% | +2.88% | 5,809 |
| Equal margin | 45.55% | +120 / 2.20 | +6.83% | +2.48% | 7,815 |
| Power | 45.46% | +120 / 2.20 | +6.82% | +2.28% | 9,241 |
| Raw, no de-vig | 47.62% | +110 / 2.10 | +7.14% | not defined | not defined |
The sample size is (z times the standard deviation, over the edge) squared: the point at which a run that returns exactly its expected value would clear the confidence level. Half of real runs come in under their expectation, so it is the number of bets before the edge could show, not the number before it reliably does. Doubling it is the honest planning figure. The bottom row is what the rest of the category reports: your price against the raw closing price, with the book's margin left inside it. It cannot produce an expected return, because the probability it implies sums to more than one across the market. The value of beating the close is the fair probability times the price difference, which is exactly zero when you bet the closing number and negative when you took a worse one.
Raw CLV vs de-vigged CLV, common price moves
Every raw CLV tracker reports the left column. The right column is what the same move is worth once the closing market's own margin comes out, computed the same way the calculator above does it, under the proportional method.
| Price move | Raw CLV | Fair expected return |
|---|---|---|
| +125 to +110 | +7.14% | +2.88% |
| -105 to -110 | +2.27% | -0.11% |
| +240 to +220 | +6.25% | +2.68% |
| -140 to -125 | -4.76% | -8.72% |
In every row the fair figure is smaller than the raw one, by roughly the hold on that closing market. That gap is not noise or a rounding artifact: it is the book's margin, counted twice by any tracker that only reports the left column.
Is landing exactly on the closing price break-even?
No. Take a two-way market priced 1.91 decimal both sides and bet the closing number exactly: raw closing line value is +0.00%, which is what a raw tracker would score as a wash. The fair expected return is -2.33%, because the closing price still carries the 2.33% hold that market has priced in. Landing exactly on the close is a small loss, not a wash, and a bettor who reliably does it is losing the margin every time, not breaking even.
Sample size before CLV means anything
At even money, with no hold on either side, so the only variable left is the edge itself. This is the two-sided sample size at 95% confidence, before any real numbers are entered below.
| Fair expected return | Bets needed |
|---|---|
| +1% | 39,189 |
| +2% | 9,993 |
| +3% | 4,529 |
| +5% | 1,695 |
Real edges after the vig comes out tend to sit in the 1% to 3% range, which is why the sample size a single season of bets can support is smaller than most bettors assume. For a bets-needed figure at your own price and edge, rather than the even-money illustration above, use the bet sample size calculator.
What JMM thinks
Closing line value is the fastest feedback a bettor has, and it is worth more than results over any sample under a few thousand bets. Almost everyone measures it against a vigged number, which flatters every record: the correction is not small, and it typically halves the measured edge. Expect the honest figure to run one or two percent rather than seven, and do not size off it until the sample is real.
The three de-vig methods on this page disagree, and the disagreement is real rather than a bug: each assumes a different rule for how a book spread its margin, and nothing in a posted price proves which rule the book used. The full comparison, with the working shown, is on the no-vig calculator. This page also assumes the closing market was efficient, which is less true at a soft book or on a market that never traded size, and it ignores limits, commission, and tax. Once an edge survives that scrutiny, the sizing question is on the Kelly calculator, and the position there is the same as it is here: assume you are measuring less edge than you think.
The next question after this one
Track record significance calculatorCalculatorCLV is the fast signal. This is the slow one: whether the actual win rate beats break-even by more than chance.Open nextOther questions
All 130 calculatorsHow JMM sources and checks its numbersWhat stays free, and what early access would add
What Pro adds here
Bet-log import, closing line value by sport and market type, and limit-weighted records are Pro features. The launch list sends one email at launch.